LMCache Files

What's Changed

• [CI]: light dockerfile build bug by @sammshen in https://github.com/LMCache/LMCache/pull/1919
• [Feat]: Pipeline Parallelism by @sammshen in https://github.com/LMCache/LMCache/pull/1813
• [core]add error handling in kv loading by @ziruiliu in https://github.com/LMCache/LMCache/pull/1835
• fix to use correct mem obj in async loading by @ziruiliu in https://github.com/LMCache/LMCache/pull/1867
• fix: minor changes to make SGL+LMCache work for TP==1 by @ziqifan617 in https://github.com/LMCache/LMCache/pull/1904
• Update bug report template with latest onboarding dashboard by @kobe0938 in https://github.com/LMCache/LMCache/pull/1906
• [feature] support set list of lookup servers by @chunxiaozheng in https://github.com/LMCache/LMCache/pull/1872
• [patch]: allow 0 buffer + layerwise detection for lmserver by @sammshen in https://github.com/LMCache/LMCache/pull/1798
• [Bugfix]Fix connection is none issue while get timeout by @maobaolong in https://github.com/LMCache/LMCache/pull/1913
• [bugfix] fix plugin interpreter lookup error by @chunxiaozheng in https://github.com/LMCache/LMCache/pull/1911
• [MINOR] Use type alias for process_tokens return value by @maobaolong in https://github.com/LMCache/LMCache/pull/1894
• Add a LMCacheBypassLookupClient to support lookup without communicate to any worker by @maobaolong in https://github.com/LMCache/LMCache/pull/1733
• [Core][Bugfix]Refactor Audit connector, avoid missing apis by @maobaolong in https://github.com/LMCache/LMCache/pull/1887
• [hotfix] guard local CPU backend creation by @DongDongJu in https://github.com/LMCache/LMCache/pull/1905
• [optimize] catch exception in batched_contains by @chunxiaozheng in https://github.com/LMCache/LMCache/pull/1920
• [bugfix] Fix the lookup socket state error once timeout occurred by @maobaolong in https://github.com/LMCache/LMCache/pull/1929
• [core][bugfix] fix memory leak in batched_put by @chunxiaozheng in https://github.com/LMCache/LMCache/pull/1927
• [PATCH]: add dtype to CEK by @sammshen in https://github.com/LMCache/LMCache/pull/1859
• [Docs]: Add docs for InfiniStore storage backend by @profetia in https://github.com/LMCache/LMCache/pull/1783
• [CI/Build] Loosen threshold for local cpu and layerwise tests by @Shaoting-Feng in https://github.com/LMCache/LMCache/pull/1947
• [CI]: audit connector unit test patch by @sammshen in https://github.com/LMCache/LMCache/pull/1951
• [InternalApi] Support config import module for the run_script api by @maobaolong in https://github.com/LMCache/LMCache/pull/1948
• [Improve|Core] Add default batched_get_non_blocking and batched_async_contains by @maobaolong in https://github.com/LMCache/LMCache/pull/1870
• Add async loading event related metrics by @maobaolong in https://github.com/LMCache/LMCache/pull/1935
• [Bugfix] Fix layerwise codepath by @YaoJiayi in https://github.com/LMCache/LMCache/pull/1950
• Bump actions/download-artifact from 5.0.0 to 6.0.0 by @dependabot[bot] in https://github.com/LMCache/LMCache/pull/1890
• Bump actions/upload-artifact from 4.6.2 to 5.0.0 by @dependabot[bot] in https://github.com/LMCache/LMCache/pull/1891
• [Core] Add SageMaker HyperPod remote connector by @ningziwen in https://github.com/LMCache/LMCache/pull/1937
• refactor: Use type alias instead of tuple for ProcessedChunk/ProcessTokensInternalResult by @maobaolong in https://github.com/LMCache/LMCache/pull/1953
• [Bugfix] Fix key comparison by @YaoJiayi in https://github.com/LMCache/LMCache/pull/1954
• [metrics]filter out 0 hit rate by @chunxiaozheng in https://github.com/LMCache/LMCache/pull/1921
• [CI/Build] Add /opt/venv/bin to PATH by @ningziwen in https://github.com/LMCache/LMCache/pull/1925

New Contributors

• @ziqifan617 made their first contribution in https://github.com/LMCache/LMCache/pull/1904
• @profetia made their first contribution in https://github.com/LMCache/LMCache/pull/1783
• @ningziwen made their first contribution in https://github.com/LMCache/LMCache/pull/1937

Full Changelog: https://github.com/LMCache/LMCache/compare/v0.3.9...v0.3.9post1